laser ablation of PCB resist

[(*steve*)]

As mentioned in this thread (https://www.electronicspoint.com/home-made-pcbs-t267347p2.html) I am currently investigating using a laser cutter to ablate resist from PCB before etching.

I have created this thread by request, I won't have any real information for another 12 to 18 hours.

 

[(*steve*)]

Some more background.

I have partially prepared three board coatings. These are three enamel spray painted pieces of double-sided board.

I would have preferred an acrylic paint but the place I went to did not have acrylic paint in spray cans.

The plan is to first try tests of speed and power to determine how the paint reacts.

Then I'll use one of my test targets on a larger board. I plan on using up to three different methods:

1) simple cutting of outlines
2) "engraving" of areas to be etched
3) multiple cutting of outlines.

The laser has a kerf of under 0.1mm so the first method will not produce great separation, and the second will take a long time to do. Hopefully the third method will be a reasonable balance.

Getting the power/speed right is actually a major step. I may not get any further than this tonight.

 

[BobK]

So, the idea is the coat the board with resist and use a laser to remove the resist wherever you want the copper removed?

Bob

 

[jpanhalt]

Can you get Krylon brand pint? It is acrylic. As an alternative, do you have conformal coating? Mine is acrylic. As a last resort, consider model airplane dope. It is not acrylic, but should vaporize better than enamel.

PS: I assume you are considering the "clear" versions.

John

 

[(*steve*)]

So, the idea is the coat the board with resist and use a laser to remove the resist wherever you want the copper removed?

Yeah, sure.

Here is a kickstarter project for a ridiculously small laser cutter where they demonstrate the technique.

The laser cutter I'll be using is a "small" one with a 30W laser and a beam width of under 0.1mm

 

[(*steve*)]

Can you get Krylon brand pint? It is acrylic.

I was unable to find any yesterday and I'm on a tight time budget.

As an alternative, do you have conformal coating? Mine is acrylic. As a last resort, consider model airplane dope. It is not acrylic, but should vaporize better than enamel.

I'll keep them in mind

PS: I assume you are considering the "clear" versions.

No, black at the moment.

I want to be able to see where it has ablated.

 

[QuantumCheese]

Curious..... Surely it's going to take a while to etch with such a narrow laser, but i'm intrigued to know how it goes.
I'm wondering how this process will compare with routing the copper directly? as that requires almost all the same kit i would have thought, X -Y table etc, just swap drill for laser and play with tracking speeds?

 

[jpanhalt]

Just a clarification: By "clear" I meant not opaque. Opaque paints usually have metal oxides in them. Those oxides may not ablate with your laser, and the residue could be messy.

I did a little checking on IR absorption of the CO2 bands. Looked at 9.4 µ (1064 cm-1)and 10.6 µ (943 cm-1). Cast acrylic is relatively transparent in that area compared to nitrocellulose and similar. So, clear lacquer and dopes may be best from that standpoint. Overall, I don't think it will really matter that much as even cast acrylic has some absorption. It you use an organic dye (i.e., transparent) color, I suspect IR absorption will be a non-issue.

John

 

[(*steve*)]

I've just got back.

The good news is that the laser ablates the paint with ease (hey even the small laser cutter is 80W!) If anything the laser is too powerful.

The main drawbacks were getting the output files from the tool I was using to draw the PCB into a form suitable for accurate cutting. The other problem was the queue to use the laser cutter!

It seems clear to me that using the laser as a milling machine to cut around the tracks is better than using it in raster mode to ablate all the non-track material.

My first test was a couple of concentric squares (can squares be concentric?). When set to "cut" I accidentally replicates the lines 3 times so it went over the path 3 times. The width of the cut material seemed to be about 1/4 to 1/2 mm. When set to "engrave" the turn on/off time of the laser was such that the edges were nowhere near as sharp.

I'll take some photos later, this is a background task, I have a few other things that are more time critical...

 

[eKretz]

Concentric = sharing the same center. So yes, squares can be concentric. I am eager to see the pictures, sounds like an interesting idea.

 

[(*steve*)]

OK, here's some quick photos.

This was the first test:

Note the poor paint job. This is what happens if you ignore the maximum temperature for using paint. It dries so fast that bubbles dry in place.

Anyway, this was using an 80W laser at 25% power and 350mm/s speed. Due to an error on my part, the laser cut each line three times :-(

The squares are 3mm, 6mm, and 9mm in each side. This gives you 1.5mm between the centres of the cut lines and my estimate is that the cut lines are between 0.3mm and 0,4mm wide.

You can also see discolouration in some corners. I did not use any reduction in laser power at the corners as the laser is a bit dodgy at powers below 25%.

Here is the second test:

This is a better paint job

This is a test of using the "etch" functionality. Again the power settings were 25% and 350mm/s.

This looks really good, but takes forever (well, a lot longer anyway)

The major problem I had was converting the board layout I had to a form suitable for the laser cutter.

The third test was using a bitmap image!

The pads you see there are for 0805 components. The IC in the middle is an LM2795 a 3mm x 3mm BGA package with 14 balls (uSMD-14)

I'm actually reasonably happy with this, but I want to do this as if the laser was a milling machine.

This is also not my paint of choice, but it's actually hard to distinguish it. This is a paint used for marking roads! It's also really cheap.

I'll probably etch them all later...

The tool I use to create PCBs can generate patterns to mill the areas around the tracks. I'm working on using that. That will be my next test so I can compare all the available methods.

 

[(*steve*)]

I've just found the time to etch these boards. (You can't accuse me of being overly hasty!)

It is clear that some residue is left on the board. This is probably the vaporised paint condensing back onto the board. It forms an excellent resist!

I got no visible etching at all on the board with the yellow paint :-(

The board with the black paint didn't begin etching until I poured some Ferric Chloride onto the board. That was sufficient to disturb a very thin layer of paint residue and start the etching. During the etching I gently rubbed the board to remove more.

The board currently looks like this:

Clearly this is still under etched, but it was pretty hard to see at the time.

I will continue the etching process, this time with some gentle rubbing of the board.

The writing has almost completely come off, as has the 0.1mm track at the far right and around the perimeter. Considering the laser's kerf, this isn't surprising.

What is surprising is that the very tiny tracks under the chip have managed to remain.

For those wondering about the bright copper visible, this is probably due to some gentle rubbing after the board was removed from the etchant.

 

[(*steve*)]

After a quick brush, and a couple of minutes more etching.. Ta Da!

The text is almost all gone, but from the traces that were removed, and the traces which *just* remain, and the closest spacing that was reliable, I can get some reasonable rules of thumb for what is possible.

But first to remove the paint and make sure the copper under it hasn't been etched at all...

 

[(*steve*)]

The paint comes off immediately with acetone.

And the finished board looks like this:

There is clearly a problem with the trace going between the pads of the 0805 component, but aside from that it's a successful first trial.

Incidentally, I tried removing the paint with isoporpyl alcohol first and scrubbed it really hard. This hardly removed any paint at all. This leads me to suspect I can be a little more aggressive with the board during etching.

I am quite happy that there are no areas at all where the paint has removed itself and the copper under it seems quite undisturbed.

 

[jpanhalt]

I think if you go with a clear acrylic or nitrocellulose (clear lacquer or nitrate dope), you will have better luck. I know you want to see the pattern, but now that you know it works, what's important is what the ferric chloride sees.

You may also want to look at the Edinburgh modification of ferric chloride It simply add some citric acid and reportedly gives finer resolution. I tried it, but couldn't tell the difference. http://www.nontoxicprint.com/etchcopperandbrass.htm

John

 

[(*steve*)]

Yeah, I'd be fine with clear acrylic, but the place I went to get the paint from didn't have any in a spray can.

Much of the laser work is with acrylic sheet, so I'm quite familiar with that.

As an interesting preview of what I will post later, the yellow paint actually worked. There was a reasonable etch of the "filled" area, and an even better etch of the "cut" lines. This was despite not brushing the lines at all.

On the downside, the yellow was more porous, the "good" side showing signs of surface etching -- however it was in the etchant for far longer than the other piece. The bubbly side did not(!! surprise) show ant etching through the paint. I assume it was thick enough (even with the bubbles) to prevent that.

Given that the laser time costs me $0.10 per minute (with a minimum charge of $1) I will also need to determine the cost of this process, but I feel it won't be huge. The more complex of these boards probably took between 1 and 2 minutes. "Cutting" is faster than "etching" on the laser, and with a combination of flood filling the board and etching only where absolutely necessary, I think the laser cost will be very small (It's certainly going to cost me more in fuel to drive to where the laser cutter is located).

 

[(*steve*)]

I ran some more tests 2 nights ago using different power settings etc. Suffice to say that the results were very disappointing, but mainly because some areas were over-etched and others under etched.

What I did learn was that:

a) higher power results in a wider ablated path (I already suspected that!)

b) a surface "skin" remains with the effect being dramatically more noticeable at low power settings.

Our conjecture is that at higher power the laser ablates right the way down to the copper. At lower power it ablates to the point at which the heat can be transferred to the copper and drawn away. There is some evidence that two passes at low power are sufficient to remove the remaining thin layer without making the "cut" wider.

Currently I am using 20% power on the laser. It is not guaranteed to lase at under 25%, and the difference in actual power between 20% and 21% is huge.

The small (about 10cm^2) but very detailed board takes 3 minutes to "cut" on the laser printer, it's way more complex than any real board would be.

At power settings of around 25% it appears that the laser cut ends up removing around 0.2mm, and this may be totally fine for anything except the smallest of surface mount features.

I will probably split my tests into realistic vs maximum resolution tests. I can live with 0.2mm between tracks for almost everything.

 

[Arouse1973]

Looking pretty good Steve

 

[Fish4Fun]

*steve*

I wish I had access to a CNC laser! I direct mill a lot of prototype//one-off PCBs on my little desktop CNC router...but there are caveats to the layouts...SOIC is about as fine as I attempt....I haven't found a good way to automate gcode generation....obviously I can import image files, but what is good for photo-resist//toner-transfer is not always so good for milling. In most cases I simply "re-create" the artwork manually in a dedicated CAD/CAM application....very time consuming, but typically excellent results....

I have long considered two other approaches: 1) Direct Laser (or perhaps inkjet) Printing to the PCB 2) Adding a UV laser to my desktop X/Y stage that would allow me to expose CEM boards "one dot at a time"....I have read everything I can find on laser ablation, but I just can't justify the cost of a laser powerful enough to do the job properly. Perhaps one day solid state lasers powerful enough to ablate paint will be readily available and cheap enough for a hobbyist!

Don't know how far you plan on going with this, but if you can find a suitable solid state laser powerful enough to do the job for less than ~$250 and would be willing to develop the controller for it, I would be willing to design/build a couple of fast, high accuracy XY stages....Of course if you already have access to a laser for a couple dollars a throw, you might not be all that keen on starting such a project, lol!

Fish

 

[(*steve*)]

OK, here is a very quick and dirty photo of 2 tests of the board cut at low power. The notes give you speed (mm/s) and %power.

Sorry about the low quality this was done in poor lighting with the board resting on my knees and taken with a mobile phone!

You can get some idea of scale by looking at the size of the text. The figures denote the text size in mm.

If you look closely you'll notice that what appear to be lines may be 2 closely spaced lines.

The grid of circles are of slightly varying size and are at 2.54mm spacing between their centres. To the right of that is an SO-8, below that is 8 pads for a 1206 device that contains 4 bussed resistors. To the direct right of that is an 0402 set of pads.

The three thinks that look like flowers are a footprint for a small BGA package.

The 9 things around these three BGA footprints are something like meshed forks. Again these are done at different pad widths. In all cases but one the width of the conductor is the same as the gap between the conductors. Because I'm using a zero offset for my cut, the conductors shrink.

Around the outside is a thin track which does some zig-zaging around to see if there is any difference between horizontal and vertical, and also diagonal cutting. These actually highlight the limitations of the repeatability of this laser cutter. In some areas you can clearly see a relative shift. This may be caused by backlash in the Y direction of the laser cutter I was using. My guess is that the shift is as large as 1/2 the kerf (so about 1/16mm. This may also be exacerbated by the fact that the drawing was originally converted to HPGL which has a 0.025mm resolution (1/40mm) before being converted to a DXF for the laser cutter.

Near the bottom right corner there are a series of triangles. These are 10mm long, are 1mm at the widest and 0mm at the line. This will be a fairly sensitive test of the total effect of the laser's kerf and the undercutting during etching.

Another factor that was suggested to me yesterday was that I had only allowed about 6 hours for the paint to dry. In my earlier tests the paint had been left for several days. This is likely to affect adhesion of the paint, and also probably it's tolerance to water (given that it is a water based paint).

I am going to (as I think I have mentioned) retry this with an unused section of the board (that will have had a week to cure) at a power similar to what has been used here, but with a second pass.

It is quite interesting to see the difference between the board cut at 250mm/s and the one cut at 100mm/s. The fact that there is a difference is remarkable because both boards took exactly 3 minutes and 6 seconds to cut.

The triangles suggest to me that the kerf is a little more than 1/8 of a mm. Previous tests with other materials (to get a friction fit) gave me a figure of 0.13mm, so it's in the same ballpark.

Yeah, I'm a casual member, so it costs me $10 for a night plus 10c per minute of laser time (min charge $1). So it's pretty cheap. For 8 of these, I used just over 20 minutes of laser time. (it was 6 minutes for another method.) Oh well, may as well show you that too!

Again, sorry about the really crap photo.

It's pretty hard to make out stuff, but where it's in focus you can see how it performs. You can also see the effect of the speed at which the laser can be turned on and off. This is a test where horizontal and vertical differ significantly.

The notes are speed/power/scan interval, and the presence of the same errors in alignment again are pretty suggestive of this being an issue with the plot file rather than than laser cutter.

I hope some of this is interesting.

Oh, one more thing. There was a guy on one of the crowd funding sites building a small laser cutter for this purpose. I'll try to find the link. He was using a very low power (comparatively) laser (only a couple of watts?)

OK, this is not the one I was thinking about, but here you go: https://www.kickstarter.com/project...ice-a-mini-arduino-laser-cutter-and-eng/posts